mirror of
https://git.quad4.io/RNS-Things/MeshChatX.git
synced 2026-04-22 00:15:44 +00:00
56 lines
1.9 KiB
Plaintext
56 lines
1.9 KiB
Plaintext
# Contributing to Reticulum MeshChatX
|
|
|
|
Patches are the preferred way to contribute. Create your changes locally, export a `.patch` file, and send it over Reticulum.
|
|
|
|
## Generating a patch
|
|
|
|
1. Clone or fork the repository and make your changes on a branch.
|
|
|
|
2. Stage and commit your work:
|
|
|
|
```bash
|
|
git add -A
|
|
git commit -m "Short description of the change"
|
|
```
|
|
|
|
3. Export the commit(s) as a `.patch` file:
|
|
|
|
```bash
|
|
# Single most recent commit
|
|
git format-patch -1
|
|
|
|
# Last N commits
|
|
git format-patch -N
|
|
|
|
# All commits since a branch point
|
|
git format-patch main..HEAD
|
|
```
|
|
|
|
This produces one `.patch` file per commit (for example `0001-my-change.patch`).
|
|
|
|
## Sending the patch
|
|
|
|
Send the `.patch` file as an LXMF message over Reticulum to:
|
|
|
|
```
|
|
7cc8d66b4f6a0e0e49d34af7f6077b5a
|
|
```
|
|
|
|
You can attach the file using Sideband, Meshchat, MeshchatX, or any LXMF-capable client with attachments support. Include a brief description of what the patch does in the message body.
|
|
|
|
Lastly, be patient.
|
|
|
|
## Patch guidelines
|
|
|
|
- Keep patches focused on a single change or fix.
|
|
- Test your changes before exporting.
|
|
- No need to run linting or formatting.
|
|
- No need to add a test or run the test suite.
|
|
|
|
## Generative AI policy
|
|
|
|
You may use generative AI tools when contributing, on the condition that your setup actually supplies the model with enough context to produce sound work: relevant files, constraints, failing tests, and project conventions. Vague prompts and thin context lead to wrong or generic patches; that burden is on the contributor, not the reviewers.
|
|
|
|
We prefer models that run locally or offline when that is practical for you.
|
|
|
|
Contributions must still be yours to justify and maintain. Do not submit bulk-generated changes you have not read, understood, and tested. We are not looking for unreviewed AI output or style-only churn from tools used without engineering/architectural judgment. |